Ordinary tasks, such as handling a cup of water. may become hard challenges for the survivors of brain injury. These studies investigate how healthy subjects control the arm when it comes in contact with the environment. Understanding how the nervous system learns to control the interactions with objects is of great importance in motor rehabilitation because interactive behaviors are essential to many activities of daily living. The proposed research is divided in three specific aims: AIM 1: To investigate the control of upper extremity interactions with mechanical constraints.These studies investigate how information about the shape of a surface is acquired through touch and how shape affects the execution of hand movements over the surface. Subjects are asked to move their hand while holding a robotic device that reproduces the mechanics of surfaces with variable shape. We formulate the hypotheses that, first. in the absence of vision, repeated movements of the hand over a curved surface lead to an adaptive process and second, that adaptation reveals, through the change in movement trajectories, an automatic and implicit learning of the surface geometry. AIM 2: To understand the control of objects with internal degrees of freedom.These studies are aimed at assessing how healthy subjects learn to control objects that are not rigid, like, for instance, a bucket full of water. The experiments pursue the hypothesis that subjects develop through practice internal representation of the manipulated objects. These representations allow them to predict how objects respond to applied forces and motions. To test this hypothesis, these studies make use of virtual mass-spring systems with which subjects interact. AIM 3: To identify effective training principles for enhancing skillful interactions with objects.These studies are aimed at evaluating the efficacy of different training methods for enhancing arm movement performance during object manipulation. The studies focus on training methods that may enhance one's ability to control with their arm the behavior of objects with complex dynamics. The experiments will test the efficacy of three training methods: i) the explicit information about correct movements of the hand; ii) the presentation of isolated cues about the effectiveness of control actions, and iii) the enhancement of visual feedback on the overall behavior of a manipulated object.